Speech dataset prices
Speech dataset price by seller: each published figure and its unit
Updated
A speech dataset price only compares with another once you know what it buys: one corpus outright or a block of hours.
Describe my language data Language, kind of data, hours and consent decide which buyers would look at it.
One-off, per dataset
| Seller and dataset | Starts at | Source |
|---|---|---|
| Nexdata: multilingual read speech, on Datarade | $20,000 a purchase | Datarade, Nexdata Multilingual Read Speech Data |
| Nexdata: unscripted call centre telephony speech, on Datarade | $20,000 a purchase | Datarade, Nexdata Unscripted Call Center Telephony Speech Data |
| Nexdata: multilingual code-switching speech, on Datarade | $20,000 a purchase | Datarade, Nexdata Multilingual Code-switching Speech Data |
| Nexdata: 8kHz conversational speech, on Datarade | $100,000 a purchase | Datarade, Nexdata 8kHz Conversational Speech Data |
| Linguistic Data Consortium: CALLHOME Japanese Second Edition, licensed by a non-member, as announced on LINGUIST List | $3000 | LINGUIST List, LDC new publications announcement |
Per plan of hours
Defined.ai's Data Access Plan announcement prices its Standard Plan at 315 hours of speech data for €50,000.
The sellers that print no price
The other speech data sellers on the table quote on request or print no price, so ask each for a quote in the unit you need.
Source: Speech and language data prices
Priced by language?
One price on the table is for a single named language: the LDC's CALLHOME Japanese Second Edition, telephone conversations between native Japanese speakers. The Nexdata listings that name their scope are multilingual or code-switching sets, and Defined.ai's plan names no language, so a price for another single language is a quote to ask a seller for.